Dynamic Traffic Signal Control Methods Based On Short-Term Traffic States Prediction

At present, the central intersections of a city are often controlled by a large urban traffic signal control system, the remain intersections are still controlled independently. An efficient adaptive control method for an isolated intersection still has important significance. It can improve transport efficiency of the road network and save energy consumption.The short term traffic flow prediction methods, adaptive signal control methods for isolated intersections and arterial intersections are researched and discussed in this thesis. The main work are as follows:1. A new short-term traffic flow prediction model is proposed, which is based on the kalman filtering predictions and the similarity of the daily traffic flow. The model combined kalman filtering predictions with urban traffic history flow and used fuzzy logic to determine the parameters of the model. The purpose of the model is to solve the delay problems of the kalman filter model predicts traffic flow. Experimental results show that the new model has higher accuracy and better instantaneity compared to the kalman filtering model.2. In view of adaptive signal control methods of isolate intersections probelm. Firstly, a phase model is built based on the conflictness relation of traffic streams and graph theory, which is used branch and bound method to determine the best phase of the intersection. Secondly, proposing traffic states model which is based on histroy traffic flow is proposed to detemine traffic states by using of short term traffic flow prediction. Finally, the model-free adaptive control for an isolated intersection method is proposed. Experimental results show that the average delay of an intersection is reduced by using this method.3. Aiming at problems of controlling arterial intersections, an improved algebraic method is presented and also a smooth transition method between two traffic signal plans is proposed to overcome disturbances caused by altering traffic plans. Experimental results show that the stablity of traffic system can be significantly improved by using this method. 4. A dynamic traffic control system based on simulation is developed by Visual Studio2008. The system can interact with elements of road network in Vissim through COM server type library. Also the proposed methods in this thesis can be verified by this system, and it may further provide technical supports to the actual development of traffic management and control platform.